Harmless Error
Judges' comments in the oral argument in Sam's appeal ominously suggest that the Second Circuit may be poised to apply the wrong legal standard to reach the wrong result.
(photo by Americasroof)
The hearing on Sam’s appeal was held the first week of November, two years to the day since his conviction and a few days shy of three years since the liquidity crisis that sent FTX into a tailspin.
Only 5 percent of federal criminal convictions are reversed on appeal. Those are dismal odds if you have been sentenced to 25 years in prison for a crime you didn’t commit. But in the year since Sam’s appeal brief was filed, a number of lawyers who read the briefs and followed the case have commented that Sam’s odds were much better than that. A few of them suggested to Sam’s lawyer or others that it was one of the strongest criminal appeals they could remember. Lawyers hate to give probabilities, but when pressed, most of them put the chances of reversal in the 30 to 60 percent range. Still not great odds if you are facing 25 years in prison for a crime you didn’t commit, but a lot better than 5 percent. Sam, who has seen more than enough of the criminal justice system in action, put the odds at 3 to 5 percent. Joe and I, who have also seen more than enough of the criminal justice system in action, nonetheless went with the more optimistic range.
Criminal appeals are tough to win because of the “harmless error” standard that is applied: courts will not overturn a conviction based on legal errors made by the judge or prosecution if they conclude that absent those errors, it was overwhelmingly likely that the defendant would still have been convicted. But the very things that doomed Sam at trial—Kaplan’s egregiously biased rulings in favor of the prosecution—made it more likely that the court would conclude that if they were errors, they were not harmless.
A favorable ruling by the Court of Appeals could mean the end of this nightmare. When appeals courts overturn a conviction for legal error, they typically remand for a new trial. I don’t think the Department of Justice would have the appetite to retry Sam’s case—not just because the new administration has announced an end to ‘regulation by prosecution’ in the crypto industry, of which Sam was the prime victim, but because there’s a good chance he would be acquitted. At least some of the people who declined to testify for Sam the first time around because of fear of retaliation by the prosecution have already said they would testify this time. The cooperating witnesses whose testimony was essential to secure Sam’s conviction have very different incentives now with the threat of 50 to 110 years in prison no longer hanging over their heads, and Sam’s new defense team would be properly prepared to rip their testimony-- and the government’s expert witness and other ‘evidence’ -- to shreds. All of the customers will have been repaid (99 percent already are). And, of course, the jury would get to hear all the evidence supporting Sam’s innocence that was wrongly excluded in the first trial.
On the other hand, if Sam loses the appeal, the odds that he will be saved by the legal system go way down. We would still have some cards to play. We could try to get the Supreme Court to review the case or file a habeas petition down the road. But they are all longshots and would likely be at least a couple of years away.
It can take months or even years for appeals courts to hand down their ruling after a hearing, but judges sometimes tip their hands during oral argument. Joe and I flew into New York the night before to attend it.
By the time a criminal case gets to the Court of Appeals, most of the life has been wrung out of it. No suspense, no do-or-die performances on the stand, no Perry Mason-like revelations, no dramatic denouement when the jury announces its verdict and the defendant is either set free into the arms of his loving family or taken away by marshals in handcuffs or chains. Since criminal convictions are generally reversed only for errors of law, not mistakes as to facts, oral arguments much more resemble a law school classroom than a gladiators’ pit.
The Second Circuit Court of Appeals building is in lower Manhattan, a block from the District Court where Sam was tried. The building itself, constructed in the 1930s, perfectly mirrors the austere, lofty, untouchable official business it houses. It was designed by the same architect who designed the US Supreme Court, which it resembles on the lower floors. Massive granite steps the width of the building lead up to a grand portico with twelve four-story columns. The main entrance hall, four stories high and made entirely of marble, takes up the entire ground floor. The business part of the court—courtrooms and judicial chambers— is housed in a narrow, thirty-story tower that rises incongruously from the base. The building is easy to pick out in the lower Manhattan skyline from its gold-leafed pyramidal roof, which glints in the morning and late afternoon sun-- more befitting a Masonic temple, I always thought, than a court of law.
Forty years ago, fresh out of law school, I spent a year in that courthouse as a law clerk for Judge J. Edward Lumbard. I hadn’t been back since. Lumbard’s chambers were on the 24th floor of the tower. Our only floormates were Judge Irving Kaufman and his law clerks.
Judge Kaufman will be remembered, for as long as history remembers him, as the judge who presided over the Rosenberg trial in 1951 and sentenced Julius and Ethel Rosenberg to death. He had been a federal judge for only a year when he was assigned to the case. Some believe that it wasn’t a random assignment, that he deliberately sought it out for the notoriety it would bring him. He certainly got that, although not the kind he hoped for.
In gross violation of judicial ethics, Kaufman secretly colluded with the prosecution throughout the Rosenberg trial. In most peoples’ view, his one-sided rulings for the prosecution virtually compelled the jury to convict. But what sealed his fate in the court of public opinion was the death sentence he imposed on both Rosenbergs and his speech defending it, in which he blamed them for the Korean War and its 50,000 dead, as well as potentially millions more deaths to come from their treason. A death sentence was unprecedented in an espionage case, not to mention one in which the defendants were charged only with being couriers. By Kaufman’s own acknowledgement, Julius was the prime mover and Ethel just a helper, but in pronouncing a death sentence on her as well, Kaufman said, in effect, she should have known better. “She was a mature woman--almost three years older than her husband and almost seven years older than her younger brother. She was a full-fledged partner in this crime.” The double execution left the Rosenbergs’ two young children orphans. This too Kaufman blamed on the Rosenbergs, who, he asserted, had been willing to sacrifice even their own children to their greater love for their diabolical cause.
Many people have speculated that Kaufman, new to the bench, felt he had to prove to the Gentile New York legal establishment that he had no clannish loyalty to his fellow Jews, that they could count on him to be a reliable team player. Justice Felix Frankfurter, a different sort of exceptional Jew of the time and adamantly opposed to the death penalty, was so enraged by Kaufman’s actions that he later wrote to Learned Hand, a friend and fellow judge, that he (Frankfurter) was “mean enough” to stay on the Supreme Court long enough to prevent Kaufman from taking Frankfurter’s place in the so-called “Jewish seat” on the Court.
Kaufman spent the rest of his career trying to live down the Rosenberg case. He had to have known that it was futile. When he died in 1991, the first line of his obituary in the New York Times read, as he knew it would, “Judge Irving R. Kaufman, who gained national attention in 1951 as the judge who sentenced Julius and Ethel Rosenberg to the electric chair . . . .” Everything else was a footnote. A handful of people showed up to picket his funeral. Someone interrupted the eulogy in the synagogue, shouting out, “He murdered the Rosenbergs! Let him rot in Hell!!”
I never met Judge Kaufman the year I clerked. He was rarely around chambers and Judge Lumbard, a man of great propriety, refused to sit on the same three-judge panel with him, for reasons I can guess. But I became friends with Kaufman’s law clerks, and one day, when he wasn’t there, they took me into his chambers to show me something. Under the glass top on his imposing carved-wood desk was an index card, with a one-sentence quote that appeared to have been typed many years earlier on an old manual typewriter. I can’t remember the august person to whom the quote was attributed or the exact wording of it all these years later, but the gist of it was, “I firmly believe that Julius and Ethel Rosenberg received a fair trial.” There was a great pathos to it, I thought. Imagine the inner life that needed daily absolution for forty years and thought he could find it there. There was also more than a hint of a guilty conscience, which counts for something, after all.
I hadn’t thought about Irving Kaufman in years. Standing in the lobby of the Appeals Court building now, waiting for the elevator I had taken so many times that year to the 24th floor, I was struck by the similarities to Kaplan’s behavior in Sam’s case. Not Kaufman’s secret collusion with the prosecution—no one would accuse Kaplan of that. He didn’t need to collude secretly with them. Everything he did, he did in the open courtroom. It was his visceral hate of Sam, which he made no effort to hide, his determination to force a conviction whatever it took, and, most of all, his speech at Sam’s sentence, that were Kaufman-esque. Kaplan didn’t accuse Sam of starting a world war, but he did accuse him of seeking to destroy our financial system and our democracy. Kaplan, however, has one character trait that I don’t think Kaufman had, despite everything. He seems to take pleasure in his cruelty.
During Sam’s trial, someone suggested to me that maybe Kaplan’s extreme animus towards Sam could be explained in part by Jewish self-loathing or a need to distance himself from a fellow Jew accused of a crime. It seemed unlikely to me. The New York legal establishment now is very different from the 1950s. Kaplan’s Jewish identity wouldn’t have been noteworthy when he was appointed to the bench in 1994 and it is even less noteworthy today. Notwithstanding the anti-Semitic hate mail we continue to receive, Sam’s Jewish identity wouldn’t be noteworthy either, at least to Kaplan. I think.
The hearing on Sam’s appeal was scheduled to start at 10 am. We left for the court early to leave enough time to get through security and secure seats, retracing the route we had taken every day of the trial. We didn’t expect the gauntlet of reporters and photographers we faced daily during the trial, but we assumed there would be a sizable media presence. It turned out to be unnecessary. There was no press outside and no line to get into the courthouse.
By 9:50, a handful of reporters had shown up. One came over to ask if we’d like to say anything and politely left when we declined. The relatively small courtroom was half-empty. That seemed like an ominous sign, but on the other hand, I told myself, the court had a job to do even if the rest of the world considered the case closed.
One minute into the oral argument, one of the judges on the panel interrupted Sam’s lawyer’s description of the evidence that Kaplan had wrongfully excluded from trial to say, in effect, that even if everything she said was true, the government presented “very substantial evidence of guilt.” In short, harmless error. When it was his turn to speak, the government lawyer helpfully reminded the panel what that “very substantial evidence” consisted of: “There were four people who knew about the misappropriation of customer deposits,” he said. “Three of them testified that they conspired with Sam Bankman-Fried to do that fraudulently.” Was this, then, going to be the final vindication of the prosecution’s strategy from day one? I.e., if it could get Sam’s three co-founders to confess to wrongdoing and state that they conspired with him to do it, it wouldn’t need anything else. On the other hand, if it couldn’t get them to do it, it was unlikely to obtain a conviction, as the prosecution itself acknowledged in its letters recommending leniency for the three witnesses. With the threat of decades in prison hanging over each of them, the government was virtually guaranteed success.
The judge’s comment was a more ominous sign. But I also knew from my year clerking not to over-read it. Appellate judges don’t always read the briefs with care in advance of oral arguments—sometimes they don’t read them at all—relying instead on briefing memos prepared by law clerks like me, fresh out of law school and lacking any experience to guide their judgment. When the panel sits down to decide the case and write the opinion, if they conclude it isn’t an obvious and easy affirmance, they will likely dig in deeper. On the other hand, they read the newspapers too.
But for the record: The correct legal standard for judging “harmless error” in this case is not whether the government presented “substantial evidence of Sam’s guilt.” That would be the correct standard if Sam’s appellate lawyers had raised what is referred to as a “sufficiency” claim-- the claim that the evidence the government presented, taken at face value, was insufficient for a jury to conclude he was guilty beyond a reasonable doubt. That is an almost impossible standard to meet, given the low bar used to judge “sufficiency,”[i] and would almost certainly not be met here. The correct question for the court to ask here is whether, had the jury been allowed to hear the wrongly excluded evidence, it would have convicted him anyway.[ii] If the answer is yes, then the trial judge’s ruling excluding it, while still legal error, would be judged “harmless.” If the answer is no, the defendant is entitled to a new trial in which the jurors are allowed to hear the evidence that was wrongly excluded the first time and decide the question of guilt for themselves. Whether or not the testimony of Sam’s three co-defendants, taken alone, constituted “substantial evidence” of Sam’s guilt does not resolve that question.
Suppose Kaplan had wrongly refused to admit into evidence a letter written by one of the co-defendants stating that Sam was innocent but that their lawyer had told them that if they didn’t agree to testify against him, they could be facing as much as 25 years in prison themselves. Or suppose Kaplan wrongly barred the defense from introducing expert testimony that would have established that the loans the government was alleging to be misappropriation of customer assets were all lawful under the terms of service governing the FTX exchange. In the first case, the excluded evidence would directly impeach the testimony of his co-defendants. In the second, it would raise the possibility that Sam’s co-defendants may have wrongly concluded not just that Sam’s actions were unlawful, but that their own were as well. In either scenario, it is hard to see how an appeals court could confidently conclude that, had the jury heard the wrongfully excluded evidence, it still would have convicted Sam based on his co-defendants’ incriminating testimony.
The judicial errors raised in Sam’s appeal were not some hyper-technical procedural violations. They went to the very heart of Sam’s constitutional right to defend himself against the government’s accusations. Among the errors raised on appeal, (i) Kaplan allowed the government to assert repeatedly and falsely that FTX customers would never recover their funds and FTX was insolvent at the time of the liquidity crisis, but barred the defense from introducing evidence that would have refuted those claims; (ii) Kaplan barred the defense from introducing evidence concerning the role of legal counsel in structuring and approving the loans and other arrangements that the government was claiming were a deliberate fraud perpetrated by Sam and his co-founders, after holding an unprecedented and likely unconstitutional “preview” hearing in which Sam was essentially deposed by the prosecution; (iii) Kaplan refused to order the Debtors to release any evidence of Sam’s innocence that was under their sole control, despite considerable evidence that the Debtors functioned essentially as an arm of the prosecution, subjecting them to Brady obligations; (iv) the entire trial was tainted by Kaplan’s overwhelming bias against Sam.
Some of these errors would arguably count as constitutional errors, mandating reversal unless the government could prove they were harmless “beyond a reasonable doubt.”[iii] Others would be judged under the somewhat more lenient test for unconstitutional errors under Fed. R. Crim Pro. 52(a): a court can find that the error was harmless only if it concludes that there was “overwhelming evidence” of the defendant’s guilt. Under either standard, the government bears the burden of proof, unlike in a sufficiency claim.
Still other alleged errors-- in particular, the allegation of judicial bias and the improper delegation of prosecutorial functions to the Debtors—arguably constituted a “structural error”: one that affects the framework within which the entire trial proceeds.[iv] Such errors require automatic reversal.
And yes, after refusing to allow a defense expert to testify that the loans from FTX to Alameda were in fact lawful under governing (UK) law, Kaplan failed to give the jury any instructions on the relevant law for determining whether they were authorized or were instead a misappropriation of customer assets.[v] The jury was left to guess that for themselves. How that came to be is a complicated story, tied up in the intricate bait and switch that the prosecution and Kaplan orchestrated to deflect the jury’s attention from the charge that the funds were wrongly appropriated—the basis of the indictment and the crime that the whole world believes Sam was convicted of—to the charge that, even if the loans were all lawful, Sam misrepresented the level of risk they posed to customers and investors in various public posts.[vi] That bait and switch played a critical role, I believe, in Sam’s conviction.
It is for the judges considering Sam’s appeal to determine whether any of the above rulings or actions were legal errors. But if they were erroneous, given their obvious importance both individually and collectively to the determination of guilt, it is hard to see how a court could confidently conclude that they were harmless, without letting a new jury hear the excluded evidence and decide for themselves how to weigh it against the evidence the government presented. That’s what juries are for.
[i] If, for example, the government’s case rests on the testimony of one eyewitness who claims she saw someone who looked like the defendant running away from the scene of the crime, it meets the “sufficiency” test, notwithstanding the notoriously high rate of false eyewitness identifications.
[ii] It wasn’t clear from their questions that the panel fully understood the distinction. See transcript, pp. 19-21.
[iii] Chapman v. California, 386 US 18, 24 (1967). This standard requires the government to demonstrate that the error did not contribute to the verdict obtained-- an extremely difficult test to meet.
[iv] Neder v. US, 527 US 1 (1999).
[v] Trial transcript, pp. 3153-3154. He did instruct them that on-line “click-through” contracts like the Terms of Service are enforceable, an issue not in dispute. Id.
[vi] See “The Trial of Sam Bankman-Fried,” pp. 13, 16-18, 41-46.